Ryzen 3000 Processors Listed on Russian Retail Site

I guess we'll have to wait until CES (or right before) to get a good look at the X570 boards, huh?
 
If it's backwards compatible I wonder if that means we won't be getting a PCIE4 chipset to go with or if the chips will be able to work with both a PCIE 3 and 4 chipset
 
Pcie4 chipsets were already leaked to be released around May.

I'm betting that the 3850X and the X570 boards will be announced/(possibly)released right at/after Computex since it's at the end of May...

That'll give me time to decide on my case/PSU/Cooling solution
 
i really think 5ghz could be real. amd is not use GF for these and will be using TSMC. GF sucks and amd has never been able to hit target clocks with them. on anything.
 
assuming RU site's info is correct, is adoredtv also correct about the 3KG series and Navi... mainly navi (because that is where AMD needs to be competitive again - and I need a new vid card)?
 
assuming RU site's info is correct, is adoredtv also correct about the 3KG series and Navi... mainly navi (because that is where AMD needs to be competitive again - and I need a new vid card)?

I hope to hell that it's true. It won't be the first time AMD/ATI targeted the entry and mid level graphics market hard and lowered the price just to embarrass Nvidia. So I'm hoping for it.
 
Good grief. Looks like my son will be getting a 1700 in his PC some time this year. VR PC will get itself a 3600X and I'll "suffer" with the 2700 moving to my machine. First world problems.
 
It's more like 9% according to The Stilt's deep dive over at Anand. But it's actually a bifurcated average. On the latency-sensitive side, the difference is larger. On the throughput side, it is less.
according 3 or 4 articles I read it is between 3% and 5% identical clock on both, with a one or two of the benchmarks they used being as low as 2%. The 9% to 15% is running the higher clock speeds on the Intel part and not at identical clock speeds and/or they are not running the proper memory speeds on the Ryzen.

You also have to keep in mind, 99% of all benchmarks out there are optimized for Intel architecture.

I searched for the Anandtech deep dive article, and I can't find any that has the both chips running at identical clock speeds, with the only deep dive released in April of 2018 when the 2700x was released.

Here is Techspots comparison at identical clock speeds (4Ghz):

https://www.techspot.com/article/1616-4ghz-ryzen-2nd-gen-vs-core-8th-gen/
 
Last edited:
6C/12T @ 50W? E-boner alert! :woot:

Certainly seems doable. A Dell i7-8700 (meaning no multicore enhancements, just running at TDP) doing 12 threads of Prime95 will only use 65W. Of course, it'll only run at around 3GHz.
 
according 3 or 4 articles I read it is between 3% and 5%, with a one or two of the benchmarks they used being as low as 2%. The 9% to 15% is running the higher clock speeds on the Intel part and not at identical clock speeds.

Not correct. The Stilt's tests were conducted at identical clockspeed. To date, I am unaware of a more comprehensive deep dive into Zen IPC. Though, if you have one I would be pleased to read it. The 9% figure is an average across a variety of tests, excluding 256b workloads (which dramatically favor Intel - when you include them in the average, it is 14% - I generally do NOT include them, however).

You also have to keep in mind, 99% of all benchmarks out there are optimized for Intel architecture.

That is certainly true in some cases. To what degree software optimizations are to blame is unknown and, perhaps, unknowable since we must conduct the tests with something. In any event, it doesn't matter, as we must use the hardware we have today on the software we have today, thus the tests are legitimate.

It's worth noting that AMD made some sacrifices to dramatically increase high performance core count (at least, when Zen was released) while also lowering per core price. This allowed AMD to sell us 8 core parts at 4 core prices, but it was not without tradeoffs and sacrifices. Zen's CCX (and then multi-die) concept introduced latency challenges that lowered IPC relative to Intel. Additionally there are some architectural quirks with regards to throughput. You will notice that Zen's single core performance in Cinebench is unimpressive, but its multi core performance approaches Skylake IPC levels. There's a reason for this. Most will say AMD has a better SMT implementation, but I think it's more accurate to say that Zen leaves a little more performance on the table per core, and thus SMT has more resources to work with and adds more performance than it does with Skylake (this is an oversimplification, but this post is long enough already).
 
I spent $550 on my 9900k and will probably STILL spend $5XX on a 3800x because I'm an idiot.
Then we're all idiots and just like tech. And your statement embodies what this forum is supposed to be about.

My last AMD build was 2002, Intel ever since, but I finally reached exhaustion - I went 7700k -> 8600k to get 2 extra cores for 6, now I want 2 more cores for total 8, but a move to 9700k will cost me hundreds yet again after only six months. I'm just tired of this exercise after Z270->Z370. Intel doesn't excite me anymore.

This is finally where I part ways with Intel and get an 8+ core AMD.
 
Last edited:
but a move to 9700k will cost me a bunch because of yet again needing a new motherboard (Z390)

I plopped my 9900k in my Z370 board with no issues, why buy the new chipset? It only gets you minimal extras.
 
  • Like
Reactions: DPI
like this
Great news and very believable. AdoredTV has been very accurate lately and it seems AMD has "chosen" him as their primary leaker. Companies actually do this on purpose, since it is free marketing.
 
I sware if AMD doesn't announce some sort of Zen 2 desktop product there is going to be an internet meltdown lol. If they just announce vega 20 for machine learning and epyc 2. That will an EPYC rumor fail! Either way I am one of those who keeps expectations low to keep disappointment low.
 
Some of those Zen 2 parts look almost too good to be true. 5 to 5.1GHz boost with 12 and 16 cores is awesome for desktop.

Regarding GPU side, AMD must announce their next gen enthusiast GPU's. They are just falling far too behind Nvidia.
 
I plopped my 9900k in my Z370 board with no issues, why buy the new chipset? It only gets you minimal extras.
I assumed 9700k required Z390 based on all the complaining I saw when Z390 was announced, but I hadn't paid enough attention.

I don't think I'll drop a 9700k into my Z370i though because I had an awakening:
seeing a 1700x at Microcenter for $139 -- 8 cores for $139. That was cheap enough to get just to play around with.

More importantly it started me down a rabbit hole of researching and comparing all the AMD CPU's against Intel, even contemplating a TR build, and it feels good to be excited about CPUs again. The next CPU/MB I buy will absolutely be AMD.
 
Last edited:
But the AMD intrigue and awakening I had was seeing a 1700x at Microcenter for $139 -- 8 cores for $139. It started me down a rabbit hole of researching the shit out of AMD which I'd so long ignored, and now I know it's my next build.

I sold 5 desktops when they hit $145 on newegg at the beginning of December, it was an absolute no brainer. Ryzen is an amazing platform. Honestly aside from single core speed there isnt any reason to go with an Intel. With my 2700x system I can play destiny (uses 8 cores) and stream (uses 8 cores) with no performance issues on the game or streaming!
 
  • Like
Reactions: DPI
like this
Some of those Zen 2 parts look almost too good to be true. 5 to 5.1GHz boost with 12 and 16 cores is awesome for desktop.

Regarding GPU side, AMD must announce their next gen enthusiast GPU's. They are just falling far too behind Nvidia.

I don't know about *must*, but it would be nice. Before Turing, there was 1 performance tier AMD didn't have. Now, there are two. They have presence throughout the rest of the stack, and compelling pricing through most of that. Hell, RX 570s are cheaper than 1050tis now, yet the green cards still sell. People are *only* buying low end based off of name recognition, not actual value, at this point. There isn't much more you can do when your part has price parity or better and is 30+% faster.
 
7nm on the compute dies and 14nm on the IO/interconnect die. Looking forward to what we will get...
 
according 3 or 4 articles I read it is between 3% and 5% identical clock on both, with a one or two of the benchmarks they used being as low as 2%. The 9% to 15% is running the higher clock speeds on the Intel part and not at identical clock speeds and/or they are not running the proper memory speeds on the Ryzen.

You also have to keep in mind, 99% of all benchmarks out there are optimized for Intel architecture.

I searched for the Anandtech deep dive article, and I can't find any that has the both chips running at identical clock speeds, with the only deep dive released in April of 2018 when the 2700x was released.

Here is Techspots comparison at identical clock speeds (4Ghz):

https://www.techspot.com/article/1616-4ghz-ryzen-2nd-gen-vs-core-8th-gen/

In the gaming benchmarks the gap is still at least 10% clock for clock in the same article so there is still a ways to go.. I would've switched over to a 2700x system already if it was only 3-5%. Lets see what the 3700x can do though for the high refresh crowd ;)
 
I have a 2500k and a Haswell HTPC that could use a 3800X and 3700. I can pay for them with money but there's no way I'll be making 64 and 32 GB RAM builds like I planned to do. My organs are used and abused.

In any case, this sounds too good to be true. Absolutely incredible.
 
I have a 2500k and a Haswell HTPC that could use a 3800X and 3700. I can pay for them with money but there's no way I'll be making 64 and 32 GB RAM builds like I planned to do. My organs are used and abused.

In any case, this sounds too good to be true. Absolutely incredible.

If your organs can't take it there are other ways to help pay rather than having to resort to something as barbaric as ~money~.
 
Not correct. The Stilt's tests were conducted at identical clockspeed. To date, I am unaware of a more comprehensive deep dive into Zen IPC. Though, if you have one I would be pleased to read it. The 9% figure is an average across a variety of tests, excluding 256b workloads (which dramatically favor Intel - when you include them in the average, it is 14% - I generally do NOT include them, however).

I linked an article from techspot. The so called deep dive you keep mentioning can't be found, so how about a link.
 
In the gaming benchmarks the gap is still at least 10% clock for clock in the same article so there is still a ways to go.. I would've switched over to a 2700x system already if it was only 3-5%. Lets see what the 3700x can do though for the high refresh crowd ;)
False, go look at the article again. I think there is 1 game out of all of them that may be even close to 10% (more like 8%), the rest are 2% to 5%.
 
It depends all of the cpu are rated in TDP /Watts all you need to do is get a cooler that can dissipate it, the complexity with Threadripper cooling and the dies being across from each other was something that you need to keep in mind when you were cooling Threadripper being 180 Watt rated TDP).
These chips are still rated low enough for air cooling to do a good to decent job.
As you said that it is more likely that the combination of power delivery and VRM cooling will be stretched on some B350/B450 boards but if the board can take a TDP 105 Watt cpu now it should not pose a problem to do it also for Ryzen 3000 series.

It's not like you can't air cool a X2700, but it definitely doesn't maintain boost as well as it could be. The Vcore goes up to something stupid like 1.5V under boost, which just causes the chip to overheat after several minutes using top end air with really good case ventilation.
 
Back
Top